Improvement of the forest canopy density model based on the addition of the FCC index and the average kernel implementation

Abstract

Awareness of the trend of forest canopy density classification requires an operational exact model for forest crown classification. The preliminary challenge is the separation of the forest crown from other non-warlike vegetation coverings. In the following, previous attempts to improve the performance of the FCD model, in this study, by adding the FCC index and the kernel, improved the average performance of the FCD model. The crown classification of Hyrcanian forests based on images of 1396 Landsat 8 was selected for implementation, evaluation, validation and analysis of the results. Improving the accuracy of the model is entirely sensible and even manual interpretation confirm it. The statistical analysis of the results also indicates a 10% and 24% increase in overall accuracy and kappa coefficient of the improved model compared to the initial model. Specifically, the accuracy of these two classes in the results of the improved model is about 13% and 7%, respectively.

Keywords

• forest canopy density
• forest color composite index (fcci)
• kernel function
• landsat 8
• hyrcanian forest